The nifty app CamFind has come a long way with its artificial intelligence (AI). It uses image recognition to identify an object when you point your smartphone camera at it.
However, back in 2015, its algorithms were less advanced: The app mostly used contract workers in the Philippines to quickly type what they saw through a user’s phone camera, CamFind’s cofounder confirmed to me recently.
You would not have guessed that from a press release it put out that year which touted industry-leading “deep learning technology,” but did not mention any human labelers.
Photo: Bloomberg
The practice of hiding human input in AI systems still remains an open secret among those who work in machine learning and AI. A 2019 analysis of tech start-ups in Europe by London-based MMC Ventures even found that 40 percent of purported AI start-ups showed no evidence of actually using artificial intelligence in their products.
This so-called “AI washing” should not be surprising. Global investment into AI companies has steadily risen over the past decade and more than doubled in the past year, according to market intelligence firm PitchBook.
Calling your start-up an “AI company” can lead to a funding premium of as much as 50 percent compared with other software companies, according to the MMC Ventures analysis.
Yet ignoring the workers who power these systems is leading to unfair labor practices and skewing the public’s understanding of how machine learning actually works.
In Silicon Valley, many start-ups have succeeded by following the “fake it ’til you make it” mantra. For AI companies, hiring people to prop up algorithms can become a stopgap, which on occasion becomes permanent.
Humans have been discovered secretly transcribing receipts, setting up calendar appointments or carrying out bookkeeping services on behalf of “AI systems” that got all the credit.
In 2019, a whistle-blower lawsuit against a British firm claimed customers paid for AI software that analyzed social media while staff members were doing that work instead.
There is a reason this happens so often. Building AI systems requires many hours of humans training algorithms, and some companies have fallen into the gray area between training and operating.
A common explanation is that human workers are providing “validation” or “oversight” to algorithms, like a quality-control check.
However, in some cases, these workers are doing more cognitively intensive tasks because the algorithms they oversee do not work well enough on their own. That can bolster unrealistic expectations about what AI can do.
“It’s part of this quixotic dream of super-intelligence,” says Ian Hogarth, an angel investor, visiting professor at University College London and coauthor of an annual State of AI report that was released on Tuesday.
For the hidden workers, working conditions can also be “anti-human,” he says.
That can lead to inequalities and poor AI performance.
For instance, Cathy O’Neil has noted that Facebook Inc’s machine-learning algorithms do not work well enough in stopping harmful content. (I agree.) The company could double its 15,000 content moderators, as suggested by a recent academic study.
However, Facebook could also bring its existing moderators out of the shadows.
The contract workers are required to sign strict non-disclosure agreements and are not allowed to talk about their work with friends and family, says Cori Crider, the founder of tech advocacy group Foxglove Legal, which has helped several former moderators take legal action against Facebook over allegations of psychological damage.
Facebook has said content reviewers could take breaks when they needed and were not pressured to make hasty decisions.
Moderation work is mentally and emotionally exhausting, and Crider says contractors are “optimized to within an inch of their lives” with an array of targets to hit.
Keeping these workers hidden only exacerbates the problem.
A similar issue affects Amazon.com Inc’s MTurk platform, which posts small tasks for freelancers.
In their book Ghost Workers, Microsoft Corp researchers Mary Gray and Siddharth Suri say these freelancers are part of an invisible workforce labeling, editing and sorting much of what we see on the Internet.
AI does not work without these “humans in the loop,” they say, yet people are largely undervalued.
A recent paper from academics at Princeton University and Cornell University called out data-labeling companies like Scale AI Inc and Sama Inc who pay workers in Southeast Asia and sub-Saharan Africa US$8 a day. Sure, that is a living wage in those regions, but long-term it also perpetuates income inequality.
A spokeswoman for Sama said the company has helped more than 55,000 people lift themselves out of poverty, and that higher local wages could negatively impact local markets, leading to higher costs for food and housing.
Scale AI did not respond to a request for comment.
“Microwork comes with no rights, security, or routine and pays a pittance — just enough to keep a person alive yet socially paralyzed,” writes Phil Jones, a researcher for the British employment think tank Autonomy, adding that it is disingenuous to paint such work as beneficial to a person’s skills.
Data labeling is so monotonous that Finland has outsourced it to prisoners.
Improving the employment status of these workers would make their lives better and also improve AI’s development, since feeding algorithms with inconsistent data can hurt future performance.
Crider says Facebook needs to make its content moderators full-time staff if it really wants to fix its content problems (most of them work for agencies like Accenture PLC).
The Princeton and Cornell researchers say labelers need a more visible role in the development of AI and more equitable pay.
One glimmer in the darkness: Freelancers who do microtasks on Amazon’s MTurk platform have recently been holding worker forums to approach Amazon on issues like rejected work, according to one of their representatives.
They are not creating a union per se, but their work is a unique attempt at organizing, giving AI’s ghost workers a voice they have not had until now. Here is hoping the idea catches on more broadly.
Parmy Olson is a Bloomberg Opinion columnist covering technology. She previously reported for the Wall Street Journal and Forbes and is the author of We Are Anonymous.
This column does not necessarily reflect the opinion ofthe editorial board or Bloomberg LP and its owners.
DECOUPLING? In a sign of deeper US-China technology decoupling, Apple has held initial talks about using Baidu’s generative AI technology in its iPhones, the Wall Street Journal said China has introduced guidelines to phase out US microprocessors from Intel Corp and Advanced Micro Devices Inc (AMD) from government PCs and servers, the Financial Times reported yesterday. The procurement guidance also seeks to sideline Microsoft Corp’s Windows operating system and foreign-made database software in favor of domestic options, the report said. Chinese officials have begun following the guidelines, which were unveiled in December last year, the report said. They order government agencies above the township level to include criteria requiring “safe and reliable” processors and operating systems when making purchases, the newspaper said. The US has been aiming to boost domestic semiconductor
Nvidia Corp earned its US$2.2 trillion market cap by producing artificial intelligence (AI) chips that have become the lifeblood powering the new era of generative AI developers from start-ups to Microsoft Corp, OpenAI and Google parent Alphabet Inc. Almost as important to its hardware is the company’s nearly 20 years’ worth of computer code, which helps make competition with the company nearly impossible. More than 4 million global developers rely on Nvidia’s CUDA software platform to build AI and other apps. Now a coalition of tech companies that includes Qualcomm Inc, Google and Intel Corp plans to loosen Nvidia’s chokehold by going
ENERGY IMPACT: The electricity rate hike is expected to add about NT$4 billion to TSMC’s electricity bill a year and cut its annual earnings per share by about NT$0.154 Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) has left its long-term gross margin target unchanged despite the government deciding on Friday to raise electricity rates. One of the heaviest power consuming manufacturers in Taiwan, TSMC said it always respects the government’s energy policy and would continue to operate its fabs by making efforts in energy conservation. The chipmaker said it has left a long-term goal of more than 53 percent in gross margin unchanged. The Ministry of Economic Affairs concluded a power rate evaluation meeting on Friday, announcing electricity tariffs would go up by 11 percent on average to about NT$3.4518 per kilowatt-hour (kWh)
OPENING ADDRESS: The CEO is to give a speech on the future of high-performance computing and artificial intelligence at the trade show’s opening on June 3, TAITRA said Advanced Micro Devices Inc (AMD) chairperson and chief executive officer Lisa Su (蘇姿丰) is to deliver the opening keynote speech at Computex Taipei this year, the event’s organizer said in a statement yesterday. Su is to give a speech on the future of high-performance computing (HPC) in the artificial intelligence (AI) era to open Computex, one of the world’s largest computer and technology trade events, at 9:30am on June 3, the Taiwan External Trade Development Council (TAITRA) said. Su is to explore how AMD and the company’s strategic technology partners are pushing the limits of AI and HPC, from data centers to